Using Pre-Emphasis/De-Emphasis to mimic Tape Saturation

[Passaloutre]

Looking at the Strymon White paper on tape saturation (https://www.strymon.net/studio-reel-reel-tape-dynamics-double-tracking-white-paper/) and how they incorporated it into their Deco pedal, I was struck with the idea that this maybe could be done in a DIY circuit (theirs is digital of course). They describe common techniques of tape recording to pre-emphasize low and high frequencies in the recording process and de-emphasize the same frequencies in the playback to reduce hum and hiss. Of course the tape medium acts as a limiter/clipper between the pre-emphasis and de-emphasis to result in some dynamic warming of the signal by compressing high frequency peaks (read the article if my description is inadequate).

Anyways, I was wondering if there were any other circuits out there that might replicate this, or if I can try to design a simple one. It seems this could be done with careful tailoring of the input EQ to boost the top end, then run it through the clipper of choice, and then low-pass filter it to soften those clipped highs. Could be the RIAA curve would be a suitable place to start:


I can't be the first person to think of this, so I'm curious if you guys know of any existing circuits specifically designed to do this. Or if not, I'd like to hear from some more knowledgeable than I about  how to go about designing something like this. At the most basic level, a Rangemaster with a tone control at the end ticks all the functional boxes, but I'd like to come up with something a little higher fidelity (and better tuned).

[Mark Hammer]

Pre-emphasis/de-emphasis is generally used in many BBD-based effects, like chorus and flanger pedals.  Some may use companding, but prior to the introduction of the venerable NE570/571 chips, pre-and-de was what people used, in conjunction with lowpass filtering, to keep noise levels low.  If someone wanted to mimic tape saturation, though, I think it would need to also introduce some modest clipping.

[Passaloutre]

If someone wanted to mimic tape saturation, though, I think it would need to also introduce some modest clipping.

For sure. The plan here would be to clip in-between the pre-emphasis and the de-emphasis, so that what you get out is not what you put in.

[Rob Strand]

For sure. The plan here would be to clip in-between the pre-emphasis and the de-emphasis, so that what you get out is not what you put in.

I played around with this stuff in the early 90's.   IMHO it's good for light amounts of overdrive and you get  kind of transparent sound.   The bass end is a bit tricky because if you try to restore it to the full level the sound gets muddy when you increase the overdrive.

I found the Ibanez MF5 (schematic is on the web) from the early 90's.   The Boss BD-2 also does this to some degree. Then there's the more modern versions like the Barber drive.

The MF5 restores the bass using  a peaking  boost in an attempt to clean-up the bass.

[Mark Hammer]

The Solo Dallas Schaffer Replica, that copies the guts of the wireless unit that reputedly gives Angus Young his signature sound, uses some form of companding, as did many earlier wireless units striving to keep noise to a minimum.  I have absolutely no idea how much the companding aspect, as opposed to some other part of the circuit, is responsible for the saturated tone Mr. Young is known for, but it is worth exploring.

[Rob Strand]

The Solo Dallas Schaffer Replica, that copies the guts of the wireless unit that reputedly gives Angus Young his signature sound, uses some form of companding, as did many earlier wireless units striving to keep noise to a minimum. 

The pre-emphasis/de-emphasis tape idea is more generic than the companding.

The Dolby systems were companding (and frequency).  They weren't used on all decks and didn't hit the cassette domain until about the 80's.
https://en.wikipedia.org/wiki/Dolby_noise-reduction_system

The ETI Sustain Fuzz uses a companding idea,
Page 53
http://www.americanradiohistory.com/Archive-Electronics-Today/80s/ETI-1980-10-80.pdf

Can't vouch for the distorting part.

[Keppy]

Dynacomp (or Ross compressor) has a simple pre-emphasis/de-emphasis scheme around the compression element.

[Mark Hammer]

The Solo Dallas Schaffer Replica, that copies the guts of the wireless unit that reputedly gives Angus Young his signature sound, uses some form of companding, as did many earlier wireless units striving to keep noise to a minimum. 

The pre-emphasis/de-emphasis tape idea is more generic than the companding.

The Dolby systems were companding (and frequency).  They weren't used on all decks and didn't hit the cassette domain until about the 80's.
https://en.wikipedia.org/wiki/Dolby_noise-reduction_system

The ETI Sustain Fuzz uses a companding idea,
Page 53
http://www.americanradiohistory.com/Archive-Electronics-Today/80s/ETI-1980-10-80.pdf

Can't vouch for the distorting part.

I have an NE646B Dolby B chip sitting in my parts bin.  One of these days I'm going to have to mess around with it.

[DavidRavenMoon]

The Solo Dallas Schaffer Replica, that copies the guts of the wireless unit that reputedly gives Angus Young his signature sound..

Also EVH. He used his in the studio as well.



Sent from my iPhone using Tapatalk

[ElectricDruid]

There are quite a few overdrive/distortion units that use this type of scheme. The pre-emphasis means that the harmonics and higher stuff tend to get most of the distortion, which stops the bass getting too muddy. Then the de-emphasis helps take the brittle treble fizz off the top, so it works out "warmer" overall. Definitely worth an experiment.

Tom

[Mark Hammer]

There are quite a few overdrive/distortion units that use this type of scheme. The pre-emphasis means that the harmonics and higher stuff tend to get most of the distortion, which stops the bass getting too muddy. Then the de-emphasis helps take the brittle treble fizz off the top, so it works out "warmer" overall. Definitely work an experiment.

Tom

You've pretty much described a Tube Screamer.

[ElectricDruid]

You've pretty much described a Tube Screamer.

See, told you there were quite a few!

T.

[amptramp]

If you are trying to mimic a saturation process, there must be some non-linear element in the design.  Pre-emphasis and de-emphasis on their own are linear effects that do not create the harmonics that saturation does.  Clipping via tape saturation takes place proportional to the volt-seconds on the magnetic medium so this will give you even clipping across the frequency range once pre-emphasis is applied.  The de-emphasis tends to limit the harshness of clipping since it reduces the harmonics that clipping creates.  You can have various levels of hardness of clipping where harder gives you higher harmonics but may also make the bass seem muddy.  Suppose you have a low frequency and a high frequency in the signal.  The wavefore will look like high frequencies riding on a low-frequency wave.  Hard clipping will not only clip the low waveform, it will remove the high-frequency component during the time the signal is clipped making for a muddy sound.

[teemuk]

Pre and post enhancing is employed in just about any distortion circuit that isn't supposed to sound like "fuzz". Heavy roll off of lower frequencies is a common technique to reduce obtrusive sounding, "farting" low frequency intermodulation distortion in guitar effects (they may start the roll off as early as 1kHz!). Due to lower amount of intermodulation distortion the overall distortion is less audible, more tolerable to listen, less muddy and farty. Complex chording will not turn into indecipherable clutter.

BUT do note though that in comparison to RIAA pre-enhancement applied to phonograph recordings the RIAA pre-enhancement applied to tape recordings is an INVERSE!  It doesn't roll off bass and boost treble, it does the opposite: It boosts bass and rolls off treble. In effect the lower frequencies are subject to more saturation than the higher ones. The RIAA curve for tape recording is also quite level from about 150 Hz to 1 kHz, characteristic not found from RIAA curve for phonographs (which is also the graph depicted earlier in this thread).
For reference, here is a graph depicting standard playback equalisation characteristics for tape recorders:

In practice you likely find out that the phonograph RIAA pre-equalisation results to more pleasant sounding distortion, but unfortunately that's not the way real tape saturation works. So do you want to build a distortion unit that sounds good, or do you want to build a distortion unit that mimicks tape saturation? What exactly is the goal?

By the way, I have scoped what "tape saturation" actually looks like and it's really just generic soft clipping, though slightly frequency dependent. Personally I think not much circuit ingenuity is required to replicate that particular effect (you can do this with two diodes and two resistors), and overall I think that ever since digital recording became the norm hyping this "tape saturation" stuff has mostly been blown out of proportions. It's not a magical effect of any kind, generally just plain soft clipping, and if saturation is subtle enough it compresses the dynamic range without discernibly audible effects, just like any other soft clipping. ...But if you saturate the tape enough it simply starts to sound like any other distortion unit. BTW, in my experiment I noticed that interfaces of many tape decks may also hard clip before they can actually drive a tape into complete saturation.

...Additionally, if you really – really - want to focus into finer details of "realistic" tape emulation then do note that recording process actually employs a high frequency modulation signal, "AC bias", to reduce tape distortion. Otherwise "hysteresis" would attenuate lower level signal content similarly to crossover distortion. The audio signal is therefore superimposed to high-frequency AC bias signal, which in average pushes the operation to most linear zones of tape's magnetic transfer function. Varying the bias amplitude has effect to both overall distortion and high frequency response.

So when one saturates a magnetic tape, what is clipping is actually the HF carrier wave with the superimposed signal. In practice clipping distortion of a signal recorded on a saturated tape is therefore a result of high frequency intermodulation. Clipping distortion is at highest magnitude when the carrier wave "peaks" while clipping distortion is reduced towards lower amplitudes of the carrier wave. Frequency, however, is so high that the modulation is virtually inaudible. Do you start to see why the tape bias matters regarding overall distortion?

In finer technical details you would also add effects like "tape flutter" to the emulation, to gradually shift up and down the frequency of the "AC bias" signal, which then results to different intermods and makes the distortion more "lively". One could also add dozens of other immaterial details such as "tape memory" effects, effects of different tape materials to magnetic hysteresis/saturation function, effects of magnitude of biasing, tape flux vs. pre-enhancement settings, and so on and so on. These all matter if you actually care about the little details.  But in the end what you see in scope screen when you record a sine wave signal to a saturating tape is still just generic soft clipping. I'm not too sure if we ever even heard any of those finer details unless we had some pretty messed up tape recorder, or a pretty messed up tape. Ironically, I remember when we tried to minimize all these effects when we were still forced to use magnetic tape as recording media. Now that "digital" is the new norm we, for some reason, concentrate anally to make it worse with all those little details we didn't care much about in the first place.

Do we honestly need a super realistic "tape emulation" when the sole purpose of such things is virtually just to prevent "brickwall" digital clipping by soft clipping limiting.

[Rob Strand]

So when one saturates a magnetic tape, what is clipping is actually the HF carrier wave with the superimposed signal.

Very good point!

[PRR]

> a graph depicting standard playback equalisation characteristics for tape recorders

I believe this is out of context.

It looks like a Flux correction. The usual wound tape head has a rising response. The usual tape play preamp has a falling response. For basically the same reasons as phono EQ. The disk-tape accepts only so-much wiggle or flux, but similar at all frequencies. For best use of the wax/rust we apply constant waggle/flux across the audio band. The wound-head makes more Voltage as the constant-flux vibration rate increases with frequency.



If you differentiate this curve (to match the wound-head's action) you get the curve you showed.



I *know* a tape will play on a phono EQ with mild midrange error and rising error over 5KHz (so I gave these systems cheap headphones).

*Forget* recording media EQ curves. There is huge history and debatable (NAB, IEC, Ampex, 3M) compromise in those curves, and they also correct for problems we do not have (gap-loss; though g-pickup pole size is related loss).

Run a flat signal to a clipper. You get a square wave with nasty extended harmonics.

Run a rising signal (differentiator) to a clipper and then to a roll-off (integrator). You get milder triangle waves. The high harmonics of the input are preserved because they are pre-boosted, and they don't shriek because they are rolled-off after.

You still, with any multi-tone signal (music) have glops of IM distortion, soured chords. The 19-cent hack is to pass only a narrow band which won't IM *too* bad.

A "natural" post-EQ is some approximation of normal screaming spectrum, -6dB/oct above ~1KHz. A natural shape for pre-boost of lead instruments may cut <250hz quite hard, because this is the bass/mid divide for most instrumentation.

In a different line of work, I filled a 600 seat hall with talkback with a 0.1 Watt amp (the clean mains needed >600W). I had space for a very large horn, which worked great at 600Hz and falling to 3KHz. Indeed white hiss into this came out with (soprano) speech/music-like spectrum. I rigged the mike preamp to rise over the same band, and at over a whisper the 0.1W power amp went all square. The speaker roll-off made it sound like speech. Vowels were shaved and sibilants boosted by the preamp rise, then mangled and hi-cut to give a less-annoying spectrum. Whispered speech passed flat and clean. Raised speech was distorted but highly intelligible and articulate without blasting.